Synthesis of supported intermetallic nanoparticle catalysts using ball milling: a case of CoGe on carbon black

Abstract

Intermetallic compounds often exhibit novel catalytic properties. However, their supported nanoparticles, a typical form of practical catalysts, cannot be easily synthesized due to differences in the redox potentials of components. Here, we demonstrate the synthesis of supported intermetallic catalysts via ball-milling using CoGe on carbon black as a case study.

---

Intermetallic compounds (IMCs) exhibit novel catalytic properties originating from their unique electronic and geometric structures.^1–3 Stabilization by forming compounds is also effective for suppressing the degradation due to particle aggregation, oxidation, and dissolution.^4–7 IMC catalysts are often investigated using conventional metallurgical synthesis methods, such as arc melting, followed by crushing into unsupported powder samples. IMCs can easily be prepared by metallurgical methods if they are thermodynamically stable. The obtained powders essentially possess homogeneity in terms of metallurgical phase, composition, and particle size. These advantages enable the efficient screening of catalysts and the accurate comparison of catalytic properties among IMCs.⁸

However, the particle size of crushed powder is usually in the micron range. Enlarging the surface area is necessary for practical applications. Supported nanoparticles are typically used for practical catalysts. However, the synthesis of supported IMC nanoparticles is not easy. Since they are typically prepared in liquid-phase chemical processes using metal salts as precursors, it requires significant effort to optimize the synthesis conditions to obtain small nanoparticles with a uniform distribution of multiple metals in the desired composition, without particle aggregation and element segregation, due to different redox potentials. Chemical reduction itself is difficult for highly oxidizable elements, including early transition metals. These disadvantages also result in poor reproducibility of sample quality. In addition, acidic or alkaline wastes lead to environmental impact concerns and high treatment costs.^9,10

Ball milling of metal precursors with support materials has been investigated as a simple alternative method for supported catalyst preparation.^10–13 In this context, Schüth's group has recently reported an even simpler process, consisting solely of ball milling ordinary metal powders with support materials.^14–19 In the first report, Au nanoparticles with a diameter of several nanometers were formed on α-Fe₂O₃, Co₃O₄, α-Al₂O₃, and TiO₂.¹⁴ The Au/TiO₂ catalyst was more active than Au/TiO₂ catalysts prepared by conventional impregnation methods for CO oxidation, although it was somewhat less active than the best sample in the literature. Pt, Ag, Cu, and Ni nanoparticles were also synthesized on the TiO₂ support.^14,19 This process has been extended to binary solid solution alloys: Au–Cu, Au–Pd, Au–Pt, Au–Ag, Pt–Pd, and Pd–Ag on MgO, yttria-stabilized zirconia (YSZ), MgAl₂O₄, α-Al₂O₃, and α-Fe₂O₃.^15,17,18 Using metal acetylacetonate precursors instead of pure metal powders, PtNi and PtCo ordered alloys, which transform to disordered solid solutions above the transition temperatures, were also synthesized on a carbon black support.²⁰

By these methods, IMCs in a narrow sense²¹ with no transition temperature for disordering have not been synthesized. Alloys consisting only of non-noble metals have also not been synthesized. A key advantage of IMCs is their potential to replace noble metal catalysts using only non-noble metals.^22–25 These ball-milling methods, especially those without precursors, can be suitable for synthesizing supported nanoparticle catalysts of non-noble metal IMCs, since the process is independent of redox potentials. In this study, we investigated the ball-milling synthesis of carbon-black-supported CoGe nanoparticle catalysts for selective hydrogenation of propyne as a model catalyst and reaction. Unlike many non-noble metal Fe-, Co-, and Ni-based IMCs that are ferromagnetic, CoGe is non-ferromagnetic and thus better suited for electron microscopy. CoGe exhibits relatively high propene selectivity in propyne hydrogenation. Since the active site isolation due to ordered atomic arrangements is crucial for selective hydrogenation by IMCs,^1,3 we can assess the success of the catalyst synthesis by observing enhanced activity without a loss of selectivity.

Using carbon black (CB, Vulcan XC-72R) as a support, samples were prepared via three different processes. The common conditions in all milling processes were as follows: loaded CB, 0.33 g; loaded Co/Ge ratio, 1; loaded CoGe, 5 wt%; planetary ball mill (Fritsch P-7); YSZ pot (12 mL); zirconia balls (5 mm diameter, 50 balls); ethanol as the solvent (filling the pot); and revolution speed, 300 rpm. The three different processes were as follows: (1. Arc) Co (Rare Metallic; purity > 99.9%) and Ge (Kojundo Kagaku; purity > 99.99%) pieces were alloyed by arc-melting, followed by annealing at 900 °C for 48 h in an Ar atmosphere. This ingot was crushed using an alumina pestle and mortar and sieved into <63 μm powder. This powder was milled for 48 h without CB and then for 24 h with CB. (2. Mechanical alloying: MA1) Co (Rare Metallic; purity > 99.9%, <325 mesh) and Ge (Rare Metallic; purity > 99.99%, <325 mesh) powders were milled for 48 h without CB and then for 24 h with CB. (3. MA2) Co and Ge powders and CB were simultaneously milled for 72 h. After milling in all processes, the samples with the solvent were dried in a Petri dish in air. Finally, the samples were heated under flowing H₂ at 120 °C for 0.5 h to remove adsorbed species, such as the residual solvent and water, and then annealed at 750 °C for 5 h to obtain a crystallized ordered IMC state.

Fig. 1 shows X-ray diffraction (XRD) patterns. After the milling, a broad peak was observed around the positions for the main peaks of CoGe in the Arc sample (Fig. 1a1), whereas no peak was observed aside from the broad peak of CB in the MA1 and MA2 samples (Fig. 1b1, c1). This indicates that CoGe particles in the Arc were simply crushed into small particles, whereas those in MA1 and MA2 were amorphous-like. The MA process introduces a high density of grain boundaries and large atomic displacements, the state of which is geometrically amorphous-like. Amorphous alloys can form when MA increases the free energy above that of a hypothetical amorphous state. IMCs are often amorphized by MA because disordering and slight compositional changes greatly increase the energy.²⁸ After annealing, CoGe peaks were clearly observed in the Arc sample (Fig. 1a2), whereas only relatively broad main peaks of CoGe were observed in MA1 and MA2 (Fig. 1b2, c2). This indicates that the primary particle sizes were smaller for the MA1 and MA2 samples, which were crystallized from the amorphous state, than for the Arc sample, which was top-down milled.

Fig. 1 XRD patterns of as-milled samples: (a1) “Arc as”, (b1) “MA1 as”, and (c1) “MA2 as”; and annealed samples: (a2) “Arc”, (b2) “MA1”, and (c2) “MA2”. (d1) Pattern of CB and (d2) reference pattern of CoGe^26,27 are also shown.

Fig. 2 shows representative elemental maps by scanning electron microscopy (SEM) with energy dispersive X-ray spectroscopy (EDX) for the Arc and MA1 samples. In the Arc sample, large CoGe particles were often observed (Fig. 2a) in addition to dispersed Co and Ge on CB (Fig. S1a, SI). In MA1, Co and Ge atoms were basically homogeneously dispersed on CB (Fig. 2b), although large CoGe particles were occasionally observed (Fig. S1b, SI). The observation for MA2 was similar to that for MA1 (Fig. S1c, SI). The average chemical compositions of the nine CB particles are summarized in Table 1 (also in Table S1, SI). The Co/Ge atomic ratio was somewhat Co-rich, especially for MA1. However, the error was very large due to an error in the huge signal of C. The average weight percent of CoGe was much less than 5% in all samples. The same analysis for commercial charcoal-supported 3 wt% Cu (Sigma-Aldrich) also resulted in a much lower weight percent of Cu (0.16%) than the nominal value (Table S2, SI). Therefore, the quantitative analysis is considered difficult in these systems.

Fig. 2 SEM images and EDX mapping of component elements for (a) Arc and (b) MA1 samples.

Table 1 Average of chemical compositions estimated by SEM–EDX analysis

	Unit	C	Co	Ge
Arc	wt%	99.52 ± 0.36	0.25 ± 0.19	0.23 ± 0.17
	at%	99.91 ± 0.07	0.051 ± 0.039	0.039 ± 0.028
MA1	wt%	99.04 ± 0.54	0.52 ± 0.29	0.44 ± 0.25
	at%	99.82 ± 0.10	0.106 ± 0.060	0.074 ± 0.043
MA2	wt%	98.54 ± 1.21	0.73 ± 0.65	0.73 ± 0.57
	at%	99.73 ± 0.23	0.15 ± 0.14	0.12 ± 0.10

---

However, large CoGe particles not supported on CB were occasionally observed in all samples (Fig. S1c, SI). Therefore, it is certain that the support was not perfect. In the milling of hard and soft materials, the soft one is smeared over the hard one.^17,29 The hardness of CB is likely to be similar to that of graphite, which is much softer than most metal materials. IMCs are basically harder than pure metals. Thus, CoGe is considered to be less readily supported on CB.

Given that the CB matrix effect on EDX analysis is similar for all samples, a relative comparison is possible. The CoGe content followed the order: MA2 > MA1 > Arc. The large CoGe particles were most prominently observed in the Arc sample. The literature indicates that harder support materials act as milling media for softer metals.^17,29 When hard CoGe powder was used as the initial source, milling by CB was unlikely to occur, and the particles were exclusively milled using 5 mm Φ balls. Thus, it is reasonable that many large CoGe particles remained in the Arc sample. Since the mechanisms for forming CoGe nanoparticles in the MA1 and MA2 samples include mechanical alloying from pure metal sources, the sample quality should be different from that of the Arc sample.

When comparing Arc and MA1, the hardness and brittleness of the amorphous-like CoGe alloy are probably different from those of the crystalline CoGe IMC. Since an amorphous state is generally softer than a crystal state (e.g., glass vs. quartz), MA1 should result in a smaller particle size and a greater tendency to be supported than the Arc sample, as indicated in Table 1. When comparing MA2 and MA1, one possible reason for the larger CoGe content in MA2 is a longer milling time with CB in MA2. Another reason is that the synthesis mechanisms were different. In MA1, Co and Ge particles were mechanically alloyed and then supported on CB. In MA2, Co and Ge particles might be individually milled and supported on CB, and then mechanically alloyed, according to the literature.¹⁷

Fig. 3a1–c1 show the TEM images. Dark particles were observed on the matrix. The interplanar spacings of dark particles correspond to those of CoGe (Fig. S2, SI), proving that CoGe nanoparticles were successfully supported on CB. The most frequently observed diameter ranges were 10–15 nm for the Arc and MA1 samples and 15–20 nm for the MA2 sample, as shown in Fig. 3a2–c2. The diameter distribution was sharpest for MA1, while relatively large particles were occasionally observed for Arc and MA2. The average diameters were 18.5 nm, 14.3 nm, and 20.8 nm for Arc, MA1, and MA2, respectively. The MA1 sample was the best in terms of particle size and its distribution.

Fig. 3 TEM images and histogram of particle diameter for (a) Arc, (b) MA1, and (c) MA2 samples.

Fig. 4a shows the catalytic properties. The conversion was similar in all samples under this condition, with sample amounts of 500 mg for the unsupported CoGe powder and 50 mg for the CB-supported CoGe nanoparticles. The selectivity was improved by nanosizing and supporting, rather than merely maintaining the high selectivity of the unsupported catalyst. Although the reason for selectivity improvement remains unclear, it may arise from differences in active sites: nanoparticles expose, in addition to the most stable facets, a greater fraction of less stable facets and low-coordinated sites such as facet edges compared to bulky particles. The reaction rates per weight were much higher for the supported catalysts than for the unsupported one, as shown in Fig. 4b, since the metal amount was only 2.5 mg in the supported catalysts. Specifically, the reaction rates were 190 and 117 times higher at 50 °C and 75 °C for Arc; 276 and 178 times for MA1; and 221 and 150 times for MA2, respectively, compared to the unsupported catalyst. These results demonstrate that CoGe intermetallic compounds can be successfully converted into supported nanoparticles using ball milling, achieving improved reaction rates while retaining high selectivity.

Fig. 4 (a) C₃H₄ conversion and C₃H₆ selectivity in propyne (C₃H₄) hydrogenation; (b) reaction rate per weight (bars) and conversion (circles) at 50 °C and 75 °C. In addition to data for supported nanoparticle samples, data for unsupported CoGe powder are shown for comparison. In (b), the values of the reaction rate are also displayed on bars. The values shown on the left and right sides are those at 50 °C and 75 °C, respectively, for the unsupported catalyst.

The highest reaction rate for MA1 is likely due to the smallest particle size and the sharpest size distribution. MA2 exhibited a higher reaction rate than Arc, even though the average particle diameter was larger for MA2 than for Arc. This may be attributed to the difference in the synthesis process. In Arc, the top-down milling resulted in a certain amount of large particles (Fig. 2a). Such large particles likely caused the XRD peaks to be sharp (Fig. 1a2). They were not included in the TEM images. Thus, the total number of small particles within the average diameter range estimated by TEM was likely smaller than that for MA2. In contrast, in MA2, the bottom-up mechanical alloying resulted in broad XRD peaks (Fig. 1c2), which indicates that small particles were predominantly formed, unlike the Arc sample. This is a possible reason for the higher reaction rate of MA2 compared to the Arc sample.

In summary, the synthesis of CB-supported CoGe nanoparticle catalysts using ball-milling was attempted via three different processes. The bottom-up mechanical alloying from pure metal powders with CB was more effective than the top-down milling of CoGe intermetallic powder with CB, in terms of size homogeneity and catalytic activity. The best catalyst exhibited a 178–276 times higher reaction rate per weight and higher selectivity for C₃H₄ selective hydrogenation compared to the unsupported catalyst. This process can also be effective for developing other supported intermetallic catalysts, regardless of their redox potentials.

A. F.: investigation, data curation, formal analysis, visualization, writing – original draft. T. K.: conceptualization, methodology, data curation, formal analysis, visualization, supervision, project administration, funding acquisition, writing – original draft, writing – review & editing.

This work was supported by Iketani Science and Technology Foundation (0351082-A). MEXT Leading Initiative for Excellent Young Researchers (JPMXS0320200014) also partially supported this work. XRD, SEM–EDX, and TEM were conducted using research equipment shared in the MEXT Project for promoting public utilization of advanced research infrastructure (Program for supporting construction of core facilities) (JPMXS0441000022, JPMXS0441000023, JPMXS0441000024, and JPMXS0441000025). We thank K. Tsubota for SEM–EDX analysis on Cu/C and XRD measurement on CB.

Conflicts of interest

There are no conflicts of interest to declare.

Data availability

The data supporting this article have been included as part of the SI. See DOI: https://doi.org/10.1039/d5cc04219d

Notes and references

1. M. Armbrüster, Sci. Technol. Adv. Mater., 2020, 21, 303–322.
2. Y. Nakaya and S. Furukawa, Chem. Rev., 2023, 123, 5859–5947.
3. N. Roy, K. MacIntosh, M. Eid, G. Canning and R. M. Rioux, Chem. Sci., 2025, 16, 8611–8636.
4. T. Kojima, T. Koganezaki, S. Fujii, S. Kameoka and A.-P. Tsai, Catal. Sci. Technol., 2021, 11, 4741–4748.
5. Y. Nakaya, J. Hirayama, S. Yamazoe, K.-I. Shimizu and S. Furukawa, Nat. Commun., 2020, 11, 2838.
6. T.-W. Song, M.-X. Chen, P. Yin, L. Tong, M. Zuo, S.-Q. Chu, P. Chen and H.-W. Liang, Small, 2022, 18, e2202916.
7. S. Yu, L. Bi, X. Xie, J. Lu, A. Chen and H. Jiang, Chem. Commun., 2023, 59, 12270–12273.
8. T. Kojima, S. Kameoka and A.-P. Tsai, KONA Powder Part. J., 2021, 38, 110–121.
9. L. F. Acevedo-Córdoba, O. J. Vargas-Montañez, E. A. Velasco-Rozo, I. D. Mora-Vergara, J. N. Díaz de León, D. Pérez-Martínez, E. M. Morales-Valencia and V. G. Baldovino-Medrano, Catal. Today, 2024, 429, 114474.
10. L. Yang, Z. Pan and Z. Tian, ChemCatChem, 2024, 16, e202301519.
11. T. Ishida, N. Kinoshita, H. Okatsu, T. Akita, T. Takei and M. Haruta, Angew. Chem., Int. Ed., 2008, 47, 9265–9268.
12. A. Braga, M. Armengol-Profitós, L. Pascua-Solé, X. Vendrell, L. Soler, I. Serrano, I. J. Villar-Garcia, V. Pérez-Dieste, N. J. Divins and J. Llorca, ACS Appl. Nano Mater., 2023, 6, 7173–7185.
13. Y. Nailwal, Q. Zhang, N. Brown, Z. Alsudairy, C. Harrod, M. H. Uddin, F. Akram, J. Li, Y. Liu and X. Li, Chem. – Eur. J., 2025, 31, e202500339.
14. H. Schreyer, R. Eckert, S. Immohr, J. de Bellis, M. Felderhoff and F. Schüth, Angew. Chem., Int. Ed., 2019, 58, 11262–11265.
15. J. De Bellis, M. Felderhoff and F. Schüth, Chem. Mater., 2021, 33, 2037–2045.
16. A. P. Amrute, J. De Bellis, M. Felderhoff and F. Schüth, Chem. – Eur. J., 2021, 27, 6819–6847.
17. J. De Bellis, H. Petersen, J. Ternieden, N. Pfänder, C. Weidenthaler and F. Schüth, Angew. Chem., Int. Ed., 2022, 61, e202208016.
18. K. S. Kley, J. De Bellis and F. Schüth, Catal. Sci. Technol., 2023, 13, 119–131.
19. K. Kuutti, M. K. Ghosalya, P. Porri, J. De Bellis, P. Jokimies, H. Singh, S. Wang, G. King, J. Fernández-Catalá, F. Schüth, K. Ainassaari, M. Huuhtanen, M. Huttula, S. Urpelainen and S. Rautiainen, Catal. Sci. Technol., 2025, 15, 4143–4155.
20. A. Gunnarson, J. De Bellis, T. Imhof, N. Pfänder, M. Ledendecker and F. Schüth, Chem. Mater., 2023, 35, 2006–2015.
21. T. Kojima, S. Kameoka and A.-P. Tsai, Sci. Technol. Adv. Mater., 2019, 20, 445–455.
22. F. Studt, F. Abild-Pedersen, T. Bligaard, R. Z. Sørensen, C. H. Christensen and J. K. Nørskov, Science, 2008, 320, 1320–1322.
23. M. Armbrüster, K. Kovnir, M. Friedrich, D. Teschner, G. Wowsnick, M. Hahne, P. Gille, L. Szentmiklósi, M. Feuerbacher, M. Heggen, F. Girgsdies, D. Rosenthal, R. Schlögl and Y. Grin, Nat. Mater., 2012, 11, 690–693.
24. T. Kojima, S. Kameoka, S. Fujii, S. Ueda and A.-P. Tsai, Sci. Adv., 2018, 4, eaat6063.
25. K. Wei, L.-W. Cao, Z. Li, Y. Y. Sun and N.-T. Suen, Inorg. Chem., 2025, 64, 3151–3155.
26. National Institute for Materials Science (NIMS), AtomWork, https://crystdb.nims.go.jp, (accessed July 20, 2025).
27. H. Takizawa, T. Sato, T. Endo and M. Shimada, J. Solid State Chem., 1988, 73, 40–46.
28. C. Suryanarayana, Research, 2019, 2019, 4219812.
29. O. V. Lapshin, E. V. Boldyreva and V. V. Boldyrev, Russ. J. Inorg. Chem., 2021, 66, 433–453.

---